Need to set up Clariion FC4500 to my K370 box...HELP

[Rita C Workman]

Hope they don't move this since I know I will get more hits here..So here goes-

In middle of too many projects and needed some quick disk to help a tight situation. Friendly vendor loaned my the above disk array....I have no experience using Clariion's and of course no manuals. Figured I would just (again in deperation) 'borrow' my redundant fiber connection, an A3404A, and do a direct connection to the FC4500 disk array. We got it connected, but unfortunately it shows NO_HW when I run ioscan.
Yes I have loaded Navisphere..and it runs navicli ok.
Yes I have run insf -e to build any special files - to no avail.
The box is a K370...running 11i.

I believe I have to create some sort of agent.config file...but have no info to do this..and since the ioscan shows NO_HW it seems futile.

I am open to any suggestions....how can I make this Clariion FC4500 visible to my HPUX K370 running 11i. And yes...it's is patched.

Beginning to bite my nails...and anybody who knows me knows how I am about keeping my nails..

Rgrds,
Rita

[Ashwani Kashyap]

Hi Rita ,

I think before HP can see any volmes on the clarrion , you need to create some LUN's , ar RAID groups or STORAGE groups and bind it to a processor .

I think it can either be done through Navisphere or from the laptop that is attached to the array .

I am not sure though , I have no experince on clarrion either .

Find plenty of Clariion documentation at http://powerlink.emc.com

[Michael Tully]

Hi Rita,

Yes, you have to create a file called agent.config. On my systems that get created under /opt/Navisphere/bin. Here is a snippet from a FC4500 in my env, although not directly connected.

#######################################

LastSave 3201222049
clarDescr Navisphere Agent
clarContact Michael Tully

device auto auto

user root
user patrol@db1pc206
user leem3@AUCSleem3-LP.corp.emc.com

poll 60
eventlog 100
baud 9600

################################################

Once you you've edited the file, you'll need to restart the navisphere agent.
Another intreseting feature I found is that if you set these up manually, the directory of /etc/opt/Navisphere does not get created.
You may need to set a storage group on a workstation that controls Navisphere, but I thought this was for only a FC4700. If you need further help I can give my clariion CE a call. (even he can't hide)

HTH some...
Michael

[Michael Tully]

He wasn't hiding....

To confirm: You don't need to set up storage groups unless you have a product called 'AccessLogics' installed. One the agent is running with the correct agent information you should be able to see your LUN's.

The agent.config file I listed for you is for a fairly late version of Navisphere. If you are running an ealier version, you may need to change the line in the config file that reads:

e.g.
device auto auto
to
device c25t0d0 NAVISPHERE-1 "NAVISPHERE-1"
device c26t0d0 NAVISPHERE-2 "NAVISPHERE-2"
and so on for each path.

Cheers
Michael

[Rita C Workman]

Michael & Ashwani,

Thanks so much for the tips...
No I don't have Access Logics installed. Just the Navisphere and Navicli. Like I said this is a loaner. I am left (I believe) with doing this via the navicli commands. Something I know zip about, but am learning real quick.
The guys who came to help set it up, couldn't get to what they said they needed to do. They believe the storage groups were already set up.

I am going to try your suggestion Michael and add those lines (based on what addresses I see) to the agent.config file I found. I am also going to manually create that directory you mention.

Well it's worth a shot anyway...points will be acoming and aplenty..

Thanks,
Rita

...anybody with more tips...please just jump in here !

[Karamanolis Nikos]

Hi there,

I suppose you have a managment station connected to the FC4500 and have also instaled Navisphere agent on HP-UX servers.

If the array was configured before with another server (ip address, or host name) then you will have to reconfigure it, using the managment station, to tell it to share the LUN's with your current server.

To get more information about the array

navicli -d host_name_or_device* getall

device= the primary device filename (cXtZd0) that should ALWAYS be available, even if no LUNs are configured... (only filename, no path)

Good luck (for sure you will need it....)
:)

[Rita C Workman]

Ah Nikos, therein lies the beginning of my challenge. There is no management station connected to the FC4500. I am learning on the fly navicli commands (what a joy..) !

It is the device auto auto that seems to work in the agen.config file.
I did find that the previous owner had setup Raid groups and I see things are bound (used command navicli -h getall).

NOW if I could just get ioscan to see something instead of returning NO_HW !!!!!!!

Points will be awarded !!

Rita

[Rita C Workman]

Son of a gun....I got it to claim something. It is only showing a couple disks though and I know there are about 10-20 disks at 36gb each in there, but it is at least showing CLAIMED !!!!

Went back over the few pages of info I got from the guy who was here last night on setting this up...Adding Navisphere to your PATH,MANPATH and SHLIB_PATH. Well apparently the SHLIB_PATH didn't export right. Once I got that fixed....and ran ioscan...devices now CLAIMED.

Still got a tall tower to climb to get into the castle...but at least I'm past the moat.

Rita

[Karamanolis Nikos · Accepted solution]

Hi again...

Well done, Rite...
I completly forgot the 5 hours it took EMC engineer to figure out that the SHLIB_PATH was not right. Oh! and another thing that they told us recently. You MUST start or restart Navisphere agent while on root directory?!?!?!?!?!?!?!!?

No, no comment on that!

Oh! and a well known, documented, solved bug, took them 3 days to find out and solve... And the funny thing is that the Greek engineer was talking on the phone 2 or 3 hours with the person from EMC who actually discovered the work around for this but none of them undestood that our case was exactly that bug. We had tried things for days, until 3 days later we get an e-mail, apologizing for not realizing it earlier....
Our problem was that we could see the LUNS configured ONLY through the SP that owned them. So if one SP fails we could not use the alternate link (PVLinks) through the other SP since there was no hardware, or device file.


Anyway, here is the bug:
The following is a Primus(R) eServer solution:

ID: emc58364
Domain: emc1
Solution Class: 3.X Compatibility

Fact OS: HP-UX

Fact Product: CLARiiON FC4500

Fact Firmware: Core Software 5.32.16

Fact Firmware: Core Software 6.32.16

Fact Host Level Protection: PVLinks

Symptom HP-UX host only sees primary paths to LUNs.

Cause When Core Software is upgraded from Core Software Rex.32.14 (or less) to x.32.16, failover mode autotrespass mode is disabled.

Fix

Note: If this installation is currently using system type "A" or initiator type "A" (no autotrespass, no PVLinks capability) and is going to remain at "A", this procedure does not have to be run.

For 5.32.16 (non-Access Logix) upgrade:

1. Use navicli to check the current system type:

navicli -d systemtype

If the systemtype shows as "3," go to Step 3.

2. If the systemtype shows as "2," use navicli to to change the system type to 3 (fibre open). This will cause the array to reboot.

navicli -d systemtype -config 3

3. Use navicli to set the failovermode value to 101 (enable autotrespass):

navicli -d -failovermode 101

4. Change the system type back to 2 (autotrespass). This will cause the array to reboot.

navicli -d systemtype -config 2

5. Run the ioscan and insf commands to display the LUNs:

ioscan -fnC disk

insf -e

ioscan -fnC disk

LUNs should now be seen on both paths